AI used 200 million BTC transactions to uncover potential money laundering on the Bitcoin blockchain.

The project is an expansion of one that was started in 2019 with a dataset consisting of just 200,000 transactions. The "Elliptic2" dataset is much larger and contains 122,000 tagged "subgraphs," which are groups of related nodes and transaction chains associated with illegal behavior.

More data helps AI learn more effectively. Bitcoin transactions offer a lot of clear data for AI to use. Elliptic, in collaboration with researchers from the MIT-IBM Watson AI Lab, used transactions to classify new criminal activity and identify patterns associated with money laundering in cryptocurrencies.

In an email, Tom Robinson, co-founder of Elliptic, said that their model identified common money laundering methods in Bitcoin. The email mentioned that although cryptocurrency laundering methods will change over time, using AI and deep learning can help automatically identify new money laundering trends.

A lot of the suspicious subgraphs contained "money transition chains." This means that a user sends money to one address and then sends the remaining money to another address that they control. A peeling chain is formed by this happening repeatedly.